NG198 Evidence Review F2

FINAL Management options for moderate to severe acne – pairwise comparisons

GRADE tables for review question: What is the effectiveness and acceptability of interventions for the treatment of moderate to severe acne (side effects and participant reported improvement)?

Oral antibiotics

Table 5: Clinical evidence profile for comparison of oral antibiotic versus placebo in participants with moderate to severe acne

Quality assessment No of patients Effect Importanc Quality e No of Risk of Other Oral Placeb Relative Design Inconsistency Indirectness Imprecision Absolute studies bias considerations antibiotic o (95% CI)

Skin irritation - Minocycline versus placebo

11 randomised serious7 no serious no serious very serious8 none 5/119 1/55 RR 2.31 (0.28 to 24 more per 1000 (from 13 ⊕ΟΟΟ CRITICAL trials inconsistency indirectness (4.2%) (1.8%) 19.31) fewer to 333 more) VERY LOW

GI side effects

51,2,3,4,5 randomised very serious10 no serious very serious8 none 82/1543 41/1177 RR 1.11 (0.62 to 4 more per 1000 (from 13 ⊕ΟΟΟ CRITICAL trials serious9 indirectness (5.3%) (3.5%) 2) fewer to 35 more) VERY LOW

Thrush / candiasis - Sarecyline versus placebo

24,5 randomised serious7 no serious no serious serious11 none 4/994 0/996 Peto OR 7.44 - ⊕⊕ΟΟ CRITICAL trials inconsistency indirectness (0.4%) (0%) (1.05 to 52.86) LOW

Patient reported improvement - Tetracycline versus placebo

16 randomised very no serious no serious no serious none 21/29 1/29 RR 21 (3.02 to 690 more per 1000 (from ⊕⊕ΟΟ CRITICAL trials serious9 inconsistency indirectness imprecision (72.4%) (3.4%) 145.98) 70 more to 1000 more) LOW

CI: confidence interval; GI: gastrointestinal; POR: peto odds ratio; RR: risk ratio 1 Stewart 2006

Acne vulgaris: evidence reviews for management options for moderate to severe acne – pairwise comparisons FINAL (June 2021) 143 FINAL Management options for moderate to severe acne – pairwise comparisons

2 Dubertret 2003 3 Leyden 2018 4 Moore 2018 (SC1401) 5 Moore 2018 (SC1402) 6 Braathen 1984 7 Overall risk of bias judgement: serious risk of bias. 8 Evidence downgraded by 2 levels due to risk of very serious imprecision as 95% confidence interval crosses 2 default MIDs for dichotomous outcomes. 9 Overall risk of bias judgement: very serious risk of bias 10 Evidence downgraded by 1 level due to serious inconsistency. 11 Evidence downgraded by 1 level due to risk of serious imprecision as 95% confidence interval crosses 1 default MID for dichotomous outcomes.

Table 6: Clinical evidence profile for comparison of oral antibiotic versus oral antibiotic in participants with moderate to severe acne

Quality assessment No of patients Effect Importanc Quality e No of Risk of Other Oral Oral Relative Design Inconsistency Indirectness Imprecision Absolute studies bias considerations antibiotic antibiotic (95% CI)

GI side effects - LYME vs MINO

11 randomised very no serious no serious very serious4 none 3/56 2/53 RR 1.42 (0.25 16 more per 1000 (from ⊕ΟΟΟ CRITICAL trials serious3 inconsistency indirectness (5.4%) (3.8%) to 8.16) 28 fewer to 270 more) VERY LOW

Thrush / candiasis - TETRA versus MINO

12 randomised very no serious no serious very serious4 none 1/21 1/23 RR 1.1 (0.07 4 more per 1000 (from 40 ⊕ΟΟΟ CRITICAL trials serious3 inconsistency indirectness (4.8%) (4.3%) to 16.43) fewer to 671 more) VERY LOW

Patient reported improvement - LYME vs MINO

11 randomised very no serious no serious no serious none 53/66 53/68 RR 1.03 (0.87 23 more per 1000 (from ⊕⊕ΟΟ CRITICAL trials serious3 inconsistency indirectness imprecision (80.3%) (77.9%) to 1.23) 101 fewer to 179 more) LOW

CI: confidence interval; LYME: lymecycline; MINO: minocycline; RR: risk ratio; TETRA: tetracycline 1 Bossuyt 2003 2 Khanna 2013 3 Overall risk of bias judgement: very serious risk of bias. 4 Evidence downgraded by 2 levels due to risk of very serious imprecision as 95% confidence interval crosses 2 default MIDs for dichotomous outcomes.

Oral hormonal contraceptives and hormone-modifying agents

Acne vulgaris: evidence reviews for management options for moderate to severe acne – pairwise comparisons FINAL (June 2021) 144 FINAL Management options for moderate to severe acne – pairwise comparisons

Table 7: Clinical evidence profile for comparison of hormonal treatments versus hormonal treatments in participants with moderate to severe acne

Quality assessment No of patients Effect Importanc Quality e No of Risk of Other Hormonal Hormonal Relative Design Inconsistency Indirectness Imprecision Absolute studies bias considerations treatments treatments (95% CI)

Breast tenderness - CPA/EE (Diane 50) versus NOR/EE (Minovlar)

11 randomised very no serious no serious very none 5/26 3/24 RR 1.54 (0.41 67 more per 1000 (from ⊕ΟΟΟ CRITICAL trials serious3 inconsistency indirectness serious4 (19.2%) (12.5%) to 5.76) 74 fewer to 595 more) VERY LOW

Breast tenderness - CPA/EE (Diane 50) versus CPA/EE (high dose CPA)

11 randomised very no serious no serious very none 5/26 4/26 RR 1.25 (0.38 38 more per 1000 (from ⊕ΟΟΟ CRITICAL trials serious3 inconsistency indirectness serious4 (19.2%) (15.4%) to 4.14) 95 fewer to 483 more) VERY LOW

Sexual dysfunction - CPA/EE (Diane 50) versus CPA/EE (high dose CPA)

11 randomised very no serious no serious very none 0/26 1/24 Peto OR 0.12 37 fewer per 1000 (from ⊕ΟΟΟ CRITICAL trials serious3 inconsistency indirectness serious4 (0%) (4.2%) (0 to 6.29) 42 fewer to 220 more) VERY LOW

Sexual dysfunction - CPA/EE (Diane 50) versus NOR/EE (Minovlar)

Mood disturbance - CPA/EE (Diane 50) versus CPA/EE (high dose CPA)

11 randomised very no serious no serious very none 3/26 3/26 RR 1 (0.22 to 0 fewer per 1000 (from ⊕ΟΟΟ CRITICAL trials serious3 inconsistency indirectness serious4 (11.5%) (11.5%) 4.5) 90 fewer to 404 more) VERY LOW

Mood disturbance - CPA/EE (Diane 50) versus NOR/EE (Minovlar)

11 randomised very no serious no serious very none 3/26 3/24 RR 0.92 (0.21 10 fewer per 1000 (from ⊕ΟΟΟ CRITICAL trials serious3 inconsistency indirectness serious4 (11.5%) (12.5%) to 4.14) 99 fewer to 392 more) VERY LOW

Acne vulgaris: evidence reviews for management options for moderate to severe acne – pairwise comparisons FINAL (June 2021) 145 FINAL Management options for moderate to severe acne – pairwise comparisons

Breakthrough bleeding - CPA/EE (Diane 50) versus CPA/EE (Diane 35)

12 randomised very no serious no serious very none 4/33 2/40 RR 2.42 (0.47 71 more per 1000 (from ⊕ΟΟΟ CRITICAL trials serious3 inconsistency indirectness serious4 (12.1%) (5%) to 12.42) 26 fewer to 571 more) VERY LOW

Breakthrough bleeding - CPA/EE (Diane 50) versus NOR/EE (Minovlar)

11 randomised very no serious no serious very none 2/26 3/24 RR 0.62 (0.11 47 fewer per 1000 (from ⊕ΟΟΟ CRITICAL trials serious3 inconsistency indirectness serious4 (7.7%) (12.5%) to 3.37) 111 fewer to 296 more) VERY LOW

Breakthrough bleeding - CPA/EE (Diane 50) versus CPA/EE (high dose CPA)

11 randomised very no serious no serious very none 2/26 4/26 RR 0.5 (0.1 to 77 fewer per 1000 (from ⊕ΟΟΟ CRITICAL trials serious3 inconsistency indirectness serious4 (7.7%) (15.4%) 2.5) 138 fewer to 231 more) VERY LOW

CI: confidence interval; CPA: cyproterone acetate; EE: ethinyl estradiol; NOR: northisterone; OR: odds ratio; RR: risk ratio 1 Miller 1986 2 Fugere 1990 3 Overall risk of bias judgement: very serious risk of bias. 4 Evidence downgraded by 2 levels due to risk of very serious imprecision as 95% confidence interval crosses 2 default MIDs for dichotomous outcomes. Oral isotretinoin

Table 8: Clinical evidence profile for comparison of isotretinoin versus isotretinoin in participants with moderate to severe acne

Quality assessment No of patients Effect Importanc Quality e No of Risk of Other Isotretinoi Isotretinoi Relative Design Inconsistency Indirectness Imprecision Absolute studies bias considerations n n (95% CI)

mucosal / cutaneous changes - ISO≥120.Daily≥0.5 versus ISO<120.Daily<0.5

Acne vulgaris: evidence reviews for management options for moderate to severe acne – pairwise comparisons FINAL (June 2021) 146 FINAL Management options for moderate to severe acne – pairwise comparisons

11 randomised very no serious no serious serious7 none 46/49 18/23 RR 1.2 (0.96 to 157 more per 1000 ⊕ΟΟΟ CRITICAL trials serious6 inconsistency indirectness (93.9%) (78.3%) 1.51) (from 31 fewer to 399 VERY LOW more)

mucosal / cutaneous changes - ISO<120.Daily≥0.5 versus ISO<120.Daily<0.5

31,2,3 randomised very very serious9 no serious very serious10 none 83/106 54/89 RR 1.26 (0.72 158 more per 1000 ⊕ΟΟΟ CRITICAL trials serious6 indirectness (78.3%) (60.7%) to 2.22) (from 170 fewer to 740 VERY LOW more)

mucosal / cutaneous changes - ISO<120.Daily≥0.5 versus ISO<120.Other≥0.5

14 randomised very no serious no serious serious7 none 37/38 16/22 RR 1.34 (1.03 247 more per 1000 ⊕ΟΟΟ CRITICAL trials serious6 inconsistency indirectness (97.4%) (72.7%) to 1.74) (from 22 more to 538 VERY LOW more)

mucosal / cutaneous changes - ISO<120.Daily<0.5 versus ISO<120.Alt<0.5

15 randomised serious11 no serious no serious no serious none 115/118 111/116 RR 1.02 (0.97 19 more per 1000 (from ⊕⊕⊕Ο CRITICAL trials inconsistency indirectness imprecision (97.5%) (95.7%) to 1.07) 29 fewer to 67 more) MODERAT E

New psychiatric diagnosis - ISO<120.Daily≥0.5 versus ISO<120.Daily<0.5

12 randomised serious11 no serious no serious very serious10 none 2/30 2/36 RR 1.2 (0.18 to 11 more per 1000 (from ⊕ΟΟΟ CRITICAL trials inconsistency indirectness (6.7%) (5.6%) 8.02) 46 fewer to 390 more) VERY LOW

Change in mood - ISO<120.Daily≥0.5 versus ISO<120.Daily<0.5

13 randomised serious11 no serious no serious very serious10 none 3/30 0/30 Peto OR 7.93 - ⊕ΟΟΟ CRITICAL trials inconsistency indirectness (10%) (0%) (0.79 to 79.26) VERY LOW

Relapse - ISO≥120.Daily≥0.5 versus ISO<120.Daily<0.5

11 randomised very no serious no serious no serious none 14/95 38/92 RR 0.25 (0.1 to 310 fewer per 1000 ⊕⊕ΟΟ CRITICAL trials serious6 inconsistency indirectness imprecision (14.7%) (41.3%) 0.61) (from 161 fewer to 372 LOW fewer)

Acne vulgaris: evidence reviews for management options for moderate to severe acne – pairwise comparisons FINAL (June 2021) 147 FINAL Management options for moderate to severe acne – pairwise comparisons

Relapse - ISO<120.Daily≥0.5 versus ISO<120.Daily<0.5

11 randomised very no serious no serious serious7 none 9/46 19/46 RR 0.47 (0.24 219 fewer per 1000 ⊕ΟΟΟ CRITICAL trials serious6 inconsistency indirectness (19.6%) (41.3%) to 0.93) (from 29 fewer to 314 VERY LOW fewer)

Relapse - ISO<120.Daily≥0.5 versus ISO<120.Other≥0.5

14 randomised very no serious no serious no serious none 0/38 3/22 Peto OR 0.06 128 fewer per 1000 ⊕⊕ΟΟ CRITICAL trials serious6 inconsistency indirectness imprecision (0%) (13.6%) (0.01 to 0.65) (from 48 fewer to 135 LOW fewer)

Relapse - ISO<120.Daily<0.5 versus ISO<120.Alt<0.5

15 randomised serious11 no serious no serious very serious12 none 0/118 0/116 RD 0 (-0.02 to - ⊕ΟΟΟ CRITICAL trials inconsistency indirectness (0%) (0%) 0.02) VERY LOW

Alt: alternative; ISO: isotretinoin; POR: peto odds ratio; RD: risk difference; RR: risk ratio 1 Strauss 1984a 2 Faghihi 2017 3 Kapadia 2005 4 Akman 2007 5 Dhaked 2016 6 Overall risk of bias judgement: very serious risk of bias. 7 Evidence downgraded by 1 level due to risk of serious imprecision as 95% confidence interval crosses 1 default MID for dichotomous outcomes. 8 Evidence downgraded by 2 levels due to very serious inconsistency. 9 Evidence downgraded by 2 levels due to risk of very serious imprecision as 95% confidence interval crosses 2 default MIDs for dichotomous outcomes. 10 Overall risk of bias judgement: serious risk of bias. 11 Evidence downgraded by 2 levels due to risk of very serious imprecision due to small number of events.

Physical treatments

Acne vulgaris: evidence reviews for management options for moderate to severe acne – pairwise comparisons FINAL (June 2021) 148 FINAL Management options for moderate to severe acne – pairwise comparisons

Table 9: Clinical evidence profile for comparison of physical treatments versus inactive control in participants with moderate to severe acne

Quality assessment No of patients Effect Importanc Quality e No of Risk of Other Physical Inactive Relative Design Inconsistency Indirectness Imprecision Absolute studies bias considerations treatments control (95% CI)

Skin irritation - Photodynamic therapy

11 randomised serious2 no serious no serious very none 7/24 2/23 RR 3.35 (0.78 204 more per 1000 (from ⊕ΟΟΟ CRITICAL trials inconsistency indirectness serious3 (29.2%) (8.7%) to 14.5) 19 fewer to 1000 more) VERY LOW CI: confidence interval; RR: risk ratio 1 Chen 2015 2 Overall risk of bias judgement: serious risk of bias. 3 Evidence downgraded by 2 levels due to risk of very serious imprecision as 95% confidence interval crosses 2 default MIDs for dichotomous outcomes.

Table 10: Clinical evidence profile for comparison of physical treatments versus physical treatments in participants with moderate to severe acne

Quality assessment No of patients Effect Importanc Quality e No of Risk of Other Physical Physical Relative Design Inconsistency Indirectness Imprecision Absolute studies bias considerations treatments treatments (95% CI)

Skin irritation - 5ALA-IPL-PDT versus IPL

11 randomised very no serious no serious no serious none 21/21 20/20 RR 1 (0.91 to 0 fewer per 1000 (from ⊕⊕ΟΟ CRITICAL trials serious6 inconsistency indirectness imprecision (100%) (100%) 1.1) 90 fewer to 100 more) LOW

Skin redness - Photodynamic + photothermal therapy versus photodynamic therapy

12 randomised very no serious no serious very serious7 none 0/14 2/14 Peto OR 0.13 124 fewer per 1000 ⊕ΟΟΟ CRITICAL trials serious6 inconsistency indirectness (0%) (14.3%) (0.01 to 2.11) (from 141 fewer to 159 VERY more) LOW

Acne vulgaris: evidence reviews for management options for moderate to severe acne – pairwise comparisons FINAL (June 2021) 149 FINAL Management options for moderate to severe acne – pairwise comparisons

Skin redness - 5ALA-IPL-PDT versus IPL

11 randomised very no serious no serious very serious7 none 3/21 0/20 Peto OR 7.81 - ⊕ΟΟΟ CRITICAL trials serious6 inconsistency indirectness (14.3%) (0%) (0.77 to 79.66) VERY LOW

Skin redness - MAL-RED-PDT versus RED

13 randomised serious8 no serious no serious very serious7 none 4/100 0/53 Peto OR 4.76 - ⊕ΟΟΟ CRITICAL trials inconsistency indirectness (4%) (0%) (0.59 to 38.13) VERY LOW

Changes in pigmentation - Photodynamic + photothermal therapy versus photodynamic therapy

12 randomised very no serious no serious very serious7 none 1/14 1/14 RR 1 (0.07 to 0 fewer per 1000 (from ⊕ΟΟΟ CRITICAL trials serious6 inconsistency indirectness (7.1%) (7.1%) 14.45) 66 fewer to 961 more) VERY LOW

Changes in pigmentation - PDL versus IPL

14 randomised very no serious no serious serious9 none 3/15 0/15 Peto OR 8.57 - ⊕ΟΟΟ CRITICAL trials serious6 inconsistency indirectness (20%) (0%) (0.82 to 89.45) VERY LOW

Changes in pigmentation - MAL-RED-PDT versus RED

13 randomised serious8 no serious no serious very serious7 none 2/100 0/53 Peto OR 4.67 - ⊕ΟΟΟ CRITICAL trials inconsistency indirectness (2%) (0%) (0.25 to 86.7) VERY LOW

Changes in pigmentation - PDL versus BR-LED

14 randomised very no serious no serious serious9 none 3/15 0/15 Peto OR 8.57 - ⊕ΟΟΟ CRITICAL trials serious7 inconsistency indirectness (20%) (0%) (0.82 to 89.45) VERY LOW

Patient reported improvement - Photodynamic + photothermal therapy versus photodynamic therapy

Acne vulgaris: evidence reviews for management options for moderate to severe acne – pairwise comparisons FINAL (June 2021) 150 FINAL Management options for moderate to severe acne – pairwise comparisons

12 randomised very no serious no serious serious9 none 14/14 12/14 RR 1.16 (0.91 137 more per 1000 ⊕ΟΟΟ CRITICAL trials serious6 inconsistency indirectness (100%) (85.7%) to 1.48) (from 77 fewer to 411 VERY more) LOW

Patient reported improvement - PDL versus IPL

14 randomised very no serious no serious serious9 none 15/15 12/15 RR 1.24 (0.94 192 more per 1000 ⊕ΟΟΟ CRITICAL trials serious7 inconsistency indirectness (100%) (80%) to 1.63) (from 48 fewer to 504 VERY more) LOW

Patient reported improvement - PDL versus BR-LED

14 randomised very no serious no serious serious9 none 15/15 6/15 RR 2.38 (1.31 552 more per 1000 ⊕ΟΟΟ CRITICAL trials serious7 inconsistency indirectness (100%) (40%) to 4.34) (from 124 more to 1000 VERY more) LOW

5ALA: 5-aminolevulinic acid; BR: blue red; CI: confidence interval; IPL: intense pulsed light; LED: light emitting diode; MAL: methyl aminolevulinate; OR: odds ratio; PDL: pulsed dye laser; RR: risk ratio 1 Mei 2013 2 Kim 2017 3 Pariser 2016 4 Sami 2008 5 Overall risk of bias judgement: very serious risk of bias. 6 Evidence downgraded by 2 levels due to risk of very serious imprecision as 95% confidence interval crosses 2 default MIDs for dichotomous outcomes. 7 Overall risk of bias judgement: serious risk of bias. 8 Evidence downgraded by 1 level due to risk of serious imprecision as 95% confidence interval crosses 1 default MID for dichotomous outcomes.

Topical and oral combination interventions

Acne vulgaris: evidence reviews for management options for moderate to severe acne – pairwise comparisons FINAL (June 2021) 151 FINAL Management options for moderate to severe acne – pairwise comparisons

Table 11: Clinical evidence profile for comparison of topical+oral versus topical+oral in participants with moderate to severe acne

Quality assessment No of patients Effect Importanc Quality e No of Risk of Other Topical+ora Topical+ora Relative Design Inconsistency Indirectness Imprecision Absolute studies bias considerations l l (95% CI)

Skin irritation - BPO + CLIND-topical + DOXY versus BPO + DOXY

11 randomised very no serious no serious very none 3/32 2/28 RR 1.31 (0.24 22 more per 1000 (from 54 ⊕ΟΟΟ CRITICAL trials serious3 inconsistency indirectness serious4 (9.4%) (7.1%) to 7.3) fewer to 450 more) VERY LOW

Skin irritation - DOXY+ TRET versus AZITH + TRET

12 randomised very no serious no serious very none 3/22 4/28 RR 0.95 (0.24 7 fewer per 1000 (from 109 ⊕ΟΟΟ CRITICAL trials serious3 inconsistency indirectness serious4 (13.6%) (14.3%) to 3.83) fewer to 404 more) VERY LOW

GI side effects - BPO + CLIND-topical + DOXY versus BPO + DOXY

11 randomised very no serious no serious very none 1/33 5/33 RR 0.2 (0.02 121 fewer per 1000 (from ⊕ΟΟΟ CRITICAL trials serious3 inconsistency indirectness serious4 (3%) (15.2%) to 1.62) 148 fewer to 94 more) VERY LOW

AZITH: azithromycin; BPO: benzoyl peroxide; CI: confidence interval; CLIND: clindamycin; DOXY: doxycycline; RR: risk ratio; TRET: tretinoin 1 Del Rosso 2007 2 Parsad 2001 3 Overall risk of bias judgement: very serious risk of bias. 4 Evidence downgraded by 2 levels due to risk of very serious imprecision as 95% confidence interval crosses 2 default MIDs for dichotomous outcomes.

Topical non-retinoids and retinoids

Acne vulgaris: evidence reviews for management options for moderate to severe acne – pairwise comparisons FINAL (June 2021) 152 FINAL Management options for moderate to severe acne – pairwise comparisons

Table 12: Clinical evidence profile for comparison of topical interventions versus vehicle in participants with moderate to severe acne

Quality assessment No of patients Effect Importanc Quality e No of Risk of Other Relative Design Inconsistency Indirectness Imprecision Topical Vehicle Absolute studies bias considerations (95% CI)

Skin irritation (non-retinoids) - Erythromycin versus vehicle

11 randomised very no serious no serious very serious21 none 26/109 26/90 RR 0.83 49 fewer per ⊕ΟΟΟ CRITICAL trials serious20 inconsistency indirectness (23.9%) (28.9%) (0.52 to 1000 (from 139 VERY LOW 1.32) fewer to 92 more)

Skin irritation (non-retinoids) - Clindamycin versus vehicle

12 randomised very no serious no serious very serious21 none 2/21 0/14 Peto OR - ⊕ΟΟΟ CRITICAL trials serious20 inconsistency indirectness (9.5%) (0%) 5.57 (0.32 to VERY LOW 98.36)

Skin irritation (non-retinoids) - BPO versus vehicle

13 randomised serious22 no serious no serious very serious21 none 4/30 1/31 Peto OR 89 more per ⊕ΟΟΟ CRITICAL trials inconsistency indirectness (13.3%) (3.2%) 3.75 (0.61 to 1000 (from 13 VERY LOW 23.01) fewer to 710 more)

Skin irritation (non-retinoids) - BPO + erythromycin versus vehicle

43,4,5,6 randomised very serious23 no serious serious24 none 36/479 16/309 Peto OR 60 more per ⊕ΟΟΟ CRITICAL trials serious20 indirectness (7.5%) (5.2%) 2.15 (1.2 to 1000 (from 10 VERY LOW 3.84) more to 147 more)

Skin irritation (retinoids)

67,8,9,10,11,12 randomised very no serious no serious no serious none 244/1945 15/1733 Peto OR 45 more per ⊕⊕ΟΟ CRITICAL trials serious20 inconsistency indirectness imprecision (12.5%) (0.87%) 6.21 (4.71 to 1000 (from 32 LOW 8.18) more to 62 more)

Light sensitivity (retinoids) - Tazarotene versus vehicle

Acne vulgaris: evidence reviews for management options for moderate to severe acne – pairwise comparisons FINAL (June 2021) 153 FINAL Management options for moderate to severe acne – pairwise comparisons

Quality assessment No of patients Effect Importanc Quality e No of Risk of Other Relative Design Inconsistency Indirectness Imprecision Topical Vehicle Absolute studies bias considerations (95% CI)

27,12 randomised serious22 no serious no serious very serious21 none 2/744 0/741 Peto OR - ⊕ΟΟΟ CRITICAL trials inconsistency indirectness (0.27%) (0%) 7.36 (0.46 to VERY LOW 117.66)

Participant reported improvement - BPO versus vehicle

115 randomised serious22 no serious no serious serious24 none 211/711 18/111 RR 1.83 135 more per ⊕⊕ΟΟ CRITICAL trials inconsistency indirectness (29.7%) (16.2%) (1.18 to 1000 (from 29 LOW 2.84) more to 298 more)

Participant reported improvement - Clindamycin versus vehicle

313,15,16 randomised very very serious25 no serious serious24 none 329/858 75/237 RR 2.15 364 more per ⊕ΟΟΟ CRITICAL trials serious20 indirectness (38.3%) (31.6%) (1.08 to 1000 (from 25 VERY LOW 4.26) more to 1000 more)

Participant reported improvement - BPO + clindamycin versus vehicle

215,17 randomised serious22 serious24 no serious no serious none 453/977 119/356 RR 1.89 298 more per ⊕⊕ΟΟ CRITICAL trials indirectness imprecision (46.4%) (33.4%) (1.26 to 1000 (from 87 LOW 2.86) more to 622 more)

Participant reported improvement - BPO + adapalene versus vehicle

18 randomised serious22 no serious no serious no serious none 182/204 26/65 RR 2.23 492 more per ⊕⊕⊕Ο CRITICAL trials inconsistency indirectness imprecision (89.2%) (40%) (1.65 to 1000 (from 260 MODERAT 3.02) more to 808 E more)

Participant reported improvement - Tazarotene versus vehicle

19 randomised serious22 no serious no serious serious24 none 53/141 20/69 RR 1.3 87 more per  CRITICAL trials inconsistency indirectness (37.6%) (29%) (0.85 to 1000 (from 43 LOW 1.99) fewer to 287 more)

Acne vulgaris: evidence reviews for management options for moderate to severe acne – pairwise comparisons FINAL (June 2021) 154 FINAL Management options for moderate to severe acne – pairwise comparisons

Quality assessment No of patients Effect Importanc Quality e No of Risk of Other Relative Design Inconsistency Indirectness Imprecision Topical Vehicle Absolute studies bias considerations (95% CI)

% change in scar count from baseline – Adapalene versus vehicle

117 randomised serious19 no serious no serious serious25 none increase in increase in scars from - - ⊕⊕ΟΟ CRITICAL trials inconsistency indirectness scars from baseline by 24.8% (p=0.036) LOW baseline by 4.5%

% change in scar count from baseline – Adapalene versus vehicle

118 randomised serious19 no serious no serious serious25 none decrease in increase in scars from - - ⊕⊕ΟΟ CRITICAL trials inconsistency indirectness scars from baseline by 14.4% (p value not LOW baseline by reported, reported only that 15.5% there was a statistical difference)

BPO: benzoyl peroxide; CI: confidence interval; POR: peto odds ratio; RR: risk ratio; SD: standard deviation 1 Dobson 1980 2 Kuhlman 1986 3 Sklar 1996 4 Jones 1981 5 Jones 2002 6 Thiboutot 2002 7 Feldman 2013 (study 301) 8 Feldman 2013 (study 302) 9 Stein Gold 2016 10 Tanghetti 2019 11 Tyring 2018 12 Pariser 2005 13 Thiboutot 2008 14 Braathen 1984 15 Gratton 1982 16 Pariser 2014 17 Dreno 2017 18 Dreno 2018 19 Overall risk of bias judgement: very serious risk of bias. 20 Evidence downgraded by 2 levels due to risk of very serious imprecision as 95% confidence interval crosses 2 default MIDs for dichotomous outcomes. 21 Overall risk of bias judgement: serious risk of bias. 22 Evidence downgraded by 1 level due to serious inconsistency. Random effects model was not possible.

Acne vulgaris: evidence reviews for management options for moderate to severe acne – pairwise comparisons FINAL (June 2021) 155 FINAL Management options for moderate to severe acne – pairwise comparisons

23 Evidence downgraded by 1 level due to risk of serious imprecision as 95% confidence interval crosses 1 default MID for dichotomous outcomes. 24 Evidence downgraded by 2 levels due to very serious inconsistency. 25 Evidence downgraded by 1 level due to serious imprecision surrounding small sample size.

Table 13: Clinical evidence profile for comparison of topical interventions versus topical interventions in participants with moderate to severe acne

Quality assessment No of patients Effect Importanc Quality e No of Risk of Other Relative Design Inconsistency Indirectness Imprecision Topical Topical Absolute studies bias considerations (95% CI)

Skin irritation (non-retinoids) - BPO + ERYTH versus BPO

11 randomised serious12 no serious no serious serious13 none 13/32 4/30 RR 3.05 (1.12 to 273 more per 1000 (from ⊕⊕ΟΟ CRITICAL trials inconsistency indirectness (40.6%) (13.3%) 8.31) 16 more to 975 more) LOW

Skin irritation (topical ± retinoid) - Motretinide versus BPO

12 randomised very no serious no serious no serious none 1/15 11/15 Peto OR 0.07 682 fewer per 1000 (from ⊕⊕ΟΟ CRITICAL trials serious14 inconsistency indirectness imprecision (6.7%) (73.3%) (0.02 to 0.29) 521 fewer to 719 fewer) LOW

Skin irritation (topical ± retinoid) - BPO+CLIND+TAZ versus BPO+CLIND

13 randomised serious12 no serious no serious very serious15 none 1/20 0/20 Peto OR 7.39 - ⊕ΟΟΟ CRITICAL trials inconsistency indirectness (5%) (0%) (0.15 to 372.38) VERY LOW

Skin irritation (topical ± retinoid) - TRET + CLIND versus CLIND

34,5,6 randomised serious12 no serious no serious no serious none 41/141 15/125 Peto OR 2.58 19 more per 1000 (from 6 ⊕⊕⊕Ο CRITICAL trials inconsistency indirectness imprecision 2 7 (1.52 to 4.39) more to 40 more) MODERAT (2.9%) (1.2%) E

Skin irritation (topical ± retinoid) - TRET + CLIND versus TRET

14 randomised serious12 no serious no serious very serious15 none 0/300 2/300 Peto OR 0.13 6 fewer per 1000 (from 7 ⊕ΟΟΟ CRITICAL trials inconsistency indirectness (0%) (0.67%) (0.01 to 2.16) fewer to 8 more) VERY LOW

Skin irritation (topical ± retinoid) - TRET versus CLIND

Acne vulgaris: evidence reviews for management options for moderate to severe acne – pairwise comparisons FINAL (June 2021) 156 FINAL Management options for moderate to severe acne – pairwise comparisons

Quality assessment No of patients Effect Importanc Quality e No of Risk of Other Relative Design Inconsistency Indirectness Imprecision Topical Topical Absolute studies bias considerations (95% CI)

14 randomised serious12 no serious no serious very serious15 none 2/300 0/150 Peto OR 4.50 - ⊕ΟΟΟ CRITICAL trials inconsistency indirectness (0.67%) (0%) (0.24 to 85.33) VERY LOW

Skin irritation (topical retinoid ± other)

47,8,9,10 randomised very no serious no serious very serious15 none 21/318 24/316 RR 0.87 (0.5 to 10 fewer per 1000 (from 38 ⊕ΟΟΟ CRITICAL trials serious14 inconsistency indirectness (6.6%) (7.6%) 1.53) fewer to 40 more) VERY LOW

Patient reported improvement - BPO+MICO versus BPO

111 randomised serious12 no serious no serious no serious none 45/51 27/51 RR 1.67 (1.26 to 355 more per 1000 (from ⊕⊕⊕Ο CRITICAL trials inconsistency indirectness imprecision (88.2%) (52.9%) 2.2) 138 more to 635 more) MODERAT E BPO: benzoyl peroxide; CLIND: clindamycin; CI: confidence interval; ERYTH: erythromycin; MICO: miconazole nitrate; POR: peto odds ratio; RR: risk ratio; TAZ: tazarotene TRET: tretinoin 1 Sklar 1996 2 Lassus 1984 3 Dhawan 2013 4 Dogra 2020 5 Schmidt 2011 6 Zouboulis 2000 7 Richter 1998 8 Stein Gold 2008 9 Tanghetti 2006 10 Tanghetti 2011 11 Degreef 1982 12 Overall risk of bias judgement: serious risk of bias. 13 Evidence downgraded by 1 level due to risk of serious imprecision as 95% confidence interval crosses 1 default MID for dichotomous outcomes. 14 Overall risk of bias judgement: very serious risk of bias. 15 Evidence downgraded by 2 levels due to risk of very serious imprecision as 95% confidence interval crosses 2 default MIDs for dichotomous outcomes.

Acne vulgaris: evidence reviews for management options for moderate to severe acne – pairwise comparisons FINAL (June 2021) 157